Centralize, analyze, and automate your Amazon operations
An internal AI-powered system that manages multiple Amazon Seller Central accounts — centralizing data, generating actionable insights, and alerting on risks.
Live KPIs, revenue trends, and top products
Capabilities
What's been built
📊 Data Pipeline Setup
Automated daily fetching of Amazon account metrics via SP-API Reports, stored in structured PostgreSQL tables.
- Direct SP-API integration (no third-party middleman)
- Fetches sales traffic, inventory, orders
- Structured tables: Accounts, Metrics, SKU Metrics, Alerts
- Configurable sync schedule per report type
- Automatic polling every 2 minutes with rate limit handling
- Multi-account and multi-marketplace support
🤖 AI Layer
OpenAI API analyzes daily data and generates structured, prioritized recommendations.
- Claude (Anthropic) + OpenAI as primary AI providers with using Open Router
- Generates top inefficiencies with severity scoring in insights
- Identifies scaling opportunities by product
- Flags high-risk areas (stockouts, velocity drops loss)
- Ranks all actions by priority and confidence
- Agent for queries on account data
🔔 Notification Layer
Daily AI insights delivered to Discord, Slack, Email — with configurable schedules and channels.
- Discord integration with rich embeds
- Slack integration with channel picker
- Email (SMTP) delivery with HTML templates
- Smart Workflows: 7 pre-built daily/weekly automations
- Custom schedules: hourly, daily, weekly, monthly
📋 Feedback & Tracking
Track AI recommendations, dismiss resolved items, and convert insights into automated workflows.
- Insights dashboard with actions
- One-click "Ask AI" to dive deeper into any recommendation
- One-click "Automate" to create a monitoring workflow
Automated Daily Insights
AI-generated analysis delivered to Discord, Slack, or Email every morning — no manual review needed.
Consolidated Performance View
All accounts, marketplaces, and metrics in one live dashboard with KPIs, charts, and health scores.
Reduced Manual Monitoring
7 pre-built workflows handle low stock, dead stock, sales drops, and unshipped orders automatically.
Clean Data Foundation
Structured PostgreSQL schema ready for future automation, portfolio analytics, and custom reporting.
Platform Alignment
Requirements & Current State
Every requirement from the brief — mapped to what currently exists in the platform.
This is just a prototype — everything can be configured as per the requirements.
Architecture
How the system is built
Direct SP-API integration, PostgreSQL storage, AI analysis with Claude, and multi-channel delivery.
Data Pipeline
How seller data flows into the platform
Automatic report syncing from Amazon SP-API into structured PostgreSQL tables — powering dashboards, AI, and workflows.
AI Assistant
How the AI analyzes and responds
OpenAI API processes questions against your real seller data, generates structured recommendations, and streams results in real-time.
Features
Explore every feature
Click each tab to see the feature in detail with a live screenshot.
All accounts in one view
Real-time metrics across all your Amazon accounts and marketplaces — the single source of truth for your portfolio.
- Revenue, Units Sold, Orders, AOV as live KPIs
- Revenue trend chart with date range filtering
- Top performing products ranked by revenue
- Inventory health status overview
- Sync health across all report types
- Multi-account switching
Tech Stack
Built with modern technologies
Production-grade infrastructure designed for reliability, performance, and extensibility.
Integrations
Multi-channel delivery & connectivity
Daily AI insights and alerts delivered wherever your team works.
FAQ
Common questions
The system connects directly to Amazon's Selling Partner API (SP-API) and automatically polls for report updates every 2 minutes. Reports are downloaded, parsed, and stored in structured PostgreSQL tables. No third-party tools like Zapier or Make.com are needed.
All Amazon credentials are encrypted at rest using AES-256-GCM. Access tokens are generated on-demand and never stored. The platform uses JWT-based authentication with short-lived access tokens (15min) and HTTP-only cookies. Role-based access control ensures users only see what they're authorized to.
Yes. The platform supports multiple Amazon seller accounts and automatically detects all marketplaces (US, CA, MX, UK, DE, etc.) for each account. You can switch between accounts from the dashboard.
The platform uses OpenAI and Anthropic Claude via OpenRouter as the primary LLM providers. The AI agent is built with LangGraph and has access to 14 specialized tools via the MCP (Model Context Protocol) server for querying your seller data.
Workflow results and alerts can be delivered to Discord (rich embeds), Slack (channel picker), Email (SMTP with HTML templates), and in-app notifications. Each workflow can target multiple channels with configurable schedules.
Absolutely. Beyond the 7 default templates, you can create custom workflows with any schedule (hourly, daily, weekly, monthly), custom AI prompts, and multi-channel delivery. You can also create workflows directly from the Agent.
Sales & Traffic, Orders, FBA Inventory, Settlements, Open Listings, and Merchant Listings. Each report type has a configurable sync schedule and dedicated parser with field mapping.
Yes. The architecture is built on a monorepo with a shared package, MCP tool server (extensible with custom tools), Custom MCP server connections, and a clean PostgreSQL schema with 30+ migrations. It's designed as a foundation for future automation and portfolio analytics.
Ready to get started?
Connect your Amazon accounts and start getting AI-powered insights in minutes.